Computer architecture and process for implementing a virtual vertical perimeter framework for an overloaded CPU having multiple network interfaces

• US 8,032,658 B2
• Filed: 07/06/2004
• Issued: 10/04/2011
• Est. Priority Date: 07/06/2004
• Status: Active Grant

• Current Assignee: Oracle America, Inc. (Oracle Corporation)
• Original Assignee: Oracle America, Inc. (Oracle Corporation)
• Inventors: Tripathi, Sunay
• Primary Examiner(s): JACOBS, LASHONDA T
• Application Number:
  US10/885,999
• Publication Number:
  US 20060047856A1
• Time in Patent Office: 2,646 Days
• Field of Search: 709/223, 709/250, 709/224, 709/229
• US Class Current: 709/250
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers

Abstract

Techniques and modes of operation for using network interfaces in a single or multi-CPU environment are disclosed. The techniques provide a virtual “vertical perimeter” framework suitable for processing data through multiple network interfaces assigned to a single CPU. In this framework, techniques for installing and servicing a plurality of network interfaces with a single CPU utilizing virtual vertical perimeters and sub-queues is disclosed. A computer system including at least one CPU and more network interfaces than CPU'"'"'s is disclosed. At least two of the interfaces are assigned to an associated CPU. The computer system includes an operating system with plurality of sub-queues and an associated sub-queue controller. The sub-queues are associated with at least two network interfaces assigned to the CPU. The sub-queue controller is configured for coordinating the generation and operation of the sub-queues thereby to facilitating message traffic through the at least two network interfaces.

Claims

1. A method of installing a network interface, the method comprising:

introducing a new network interface to a computer system architecture,

wherein the computer architecture prior to introducing the new network interface comprises a plurality of central processing units (CPUs) and a plurality of existing network interfaces,

wherein the number of existing network interfaces is at least as great as the number of CPUs,

wherein each existing network interface is assigned to a designated CPU of the plurality of CPUs using a designated queue of a plurality of queues, and wherein at least one existing network interface is assigned to each CPU of the plurality of CPUs;

selecting a first queue of the plurality of queues based on the first queue comprising a least amount of data amongst the plurality of queues;

identifying, from the plurality of CPUs, a CPU corresponding to the first queue based on the first queue comprising the least amount of data;

assigning the new network interface to the CPU to generate an overloaded CPU, wherein after assigning the new network interface, the overloaded CPU comprises more than one network interface assigned to the overloaded CPU;

generating, by one of the plurality of CPUs, a plurality of sub-queues sufficient to accommodate the more than one network interface assigned to the overloaded CPU, wherein each network interface of the

overloaded CPU is associated with a separate sub-queue of the plurality of sub-queues;

assigning the plurality of sub-queues to the overloaded CPU; and

generating a sub-queue controller for coordinating the operation of the sub-queues of the overloaded CPU.

2. The method of claim 1, wherein said introducing the new network interface to the computer system architecture comprises plugging a network interface card into the computer system architecture.

3. The method of claim 1, further comprising:

identifying the first queue and a second queue of the plurality of queues based on each comprising the least amount of data amongst the plurality of queues;

analyzing a message packet processing history for the first queue and the second queue to identify the first queue as less busy than the second queue,

wherein selecting the first queue is based on the first queue comprising the least amount of data amongst the plurality of queues and being less busy than the second queue.

4. The method of claim 1, wherein generating the plurality of sub-queues for the overloaded CPU comprises:

generating a new sub-queue for servicing the new network interface and converting an existing queue that previously serviced an existing network interface to a sub-queue to continue servicing the existing network interface.

5. The method of claim 1, wherein the sub-queue controller includes:

providing an indicator for each sub-queue, wherein each indicator identifies a memory location for each sub-queue assigned to each of the network interfaces associated with the overloaded CPU; and

providing an indicator for an allocation engine, wherein the indicator identifies a memory location for the allocation engine which regulates the flow of message packets from the sub-queues to the overloaded CPU to facilitate message traffic between the network interfaces and the overloaded CPU.

9. The non-transitory computer readable media of claim 1, wherein the sub-queue controller includes:

computer program code for providing an indicator for each sub-queue, wherein each indicator identifies a memory location for each sub-queue assigned to each of the network interfaces associated with the overloaded CPU; and

computer program code for providing an indicator for an allocation engine, wherein the indicator identifies a memory location for the allocation engine which regulates the flow of message packets from the sub-queues to the overloaded CPU to facilitate message traffic between the network interfaces and the overloaded CPU.

6. A non-transitory computer readable media comprising computer program code for installing a network interface in a computer system architecture, computer readable media comprising:

computer program code for introducing a new network interface to the computer system architecture,

wherein the computer architecture prior to introducing the new network interface comprises a plurality of central processing units (CPUs) and a plurality of existing network interfaces,

wherein the number of existing network interfaces is at least as great as the number of CPUs,

wherein each existing network interface is assigned to a designated CPU of the plurality of CPUs using a designated queue of a plurality of queues, and

wherein at least one existing network interface is assigned to each CPU of the plurality of CPUs;

computer program code for selecting a first queue of the plurality of queues based on the first queue comprising a least amount of data amongst the plurality of queues;

computer program code for identifying, from the plurality of CPUs, a CPU corresponding to the first queue based on the first queue comprising the least amount of data;

computer program code for assigning the new network interface to the CPU to generate an overloaded CPU, wherein after assigning the new network interface, the overloaded CPU comprises more than one network interface assigned to the overloaded CPU;

computer program code for generating a plurality of sub-queues sufficient to accommodate the more than one network interfaces assigned to the overloaded CPU, wherein each network interface of the overloaded CPU is associated with a separate sub-queue of the plurality of sub-queues;

computer program code for assigning the plurality of sub-queues to the overloaded CPU; and

computer program code for generating a sub-queue controller for coordinating the operation of the sub-queues of the overloaded CPU.

7. The non-transitory computer readable media of claim 6, further comprising:

computer program code for identifying the first queue and a second queue of the plurality of queues based on each comprising the least amount of data amongst the plurality of queues;

computer program code for analyzing a message packet processing history for the first queue and the second queue to identity the first queue as less busy than the second queue,

wherein selecting the first queue is based on the first queue comprising the least amount of data amongst the plurality of queues and being less busy than the second queue.

8. The non-transitory computer readable media of claim 6, wherein the computer program code for generating the plurality of sub-queues for the overloaded CPU comprises:

computer program code for generating a new sub-queue for servicing the new network interface and converting an existing queue that previously serviced an existing network interface to a sub-queue to continue servicing the existing network interface.

10. A computer system comprising:

a plurality of CPUs;

a plurality of existing network interfaces,

wherein the number of existing network interfaces is at least as great as the number of CPUs,

wherein each existing network interface is assigned to a designated CPU of the plurality of CPUs using a designated queue of the plurality of queues, and

wherein at least one existing network interface is assigned to each CPU of the plurality of CPUs;

an operating system comprising;

a controller configured to;

detect a new network interface to the computer system;

select a first queue of the plurality of queues based on the first queue comprising a least amount of data amongst the plurality of queues;

identify, from the plurality of CPUs, a CPU corresponding to the first queue based on the first queue comprising the least amount of data; and

assign the new network interface to the CPU to generate an overloaded CPU, wherein after assigning the new network interface, the overloaded CPU comprises more than one network interface assigned to the overloaded CPU; and

a plurality of sub-queues, wherein each sub-queue is associated with one of the network interfaces that are assigned to the overloaded CPU, wherein each of the network interfaces that are assigned to the overloaded CPU is associated with a separate sub-queue of the plurality of sub-queues.

11. The computer system of claim 10, wherein the plurality of existing network interfaces are network interface cards (NIC).

12. The computer system of claim 10, wherein the controller allocates when message packets in each sub-queue are processed by the overloaded CPU.

13. The computer system of claim 12, wherein the controller determines a priority schedule for allocating when message packets in each sub-queue are processed by the overloaded CPU.

14. The computer system of claim 13, wherein the priority schedule for allocating when the message packets in each sub-queue are to be processed by the overloaded CPU comprises a time-based allocation.

15. The computer system of claim 13, wherein the priority schedule for allocating when the message packets in each sub-queue are to be processed by the overloaded CPU comprises a packet-based allocation dependent on the number of message packets processed for each sub-queue.

16. The computer system of claim 13, wherein the priority schedule for allocating when the message packets in each sub-queue are to be processed by the overloaded CPU is based upon the relative message processing capacities of the network interfaces associated with each sub-queue.

17. The computer system of claim 13, wherein the priority schedule for allocating when the message packets in each sub-queue are to be processed by the overloaded CPU can be adjusted by a user.

18. The computer system of claim 12, wherein the determining of the priority schedule for allocating when message packets in each sub-queue are processed by the overloaded CPU is determined by a scheduler.

19. The computer system of claim 12, wherein the controller is configured to control the rate at which data is received by a sub-queue by using the occupancy rate of each sub-queue.

20. The computer system of claim 19, wherein the controller controls the rate at which message packets are received by each sub-queue by alternatively using polling and interrupt techniques to facilitate the transmission of message packets from the assigned network interface.

21. The computer system of claim 10, wherein the controller enables dynamic installation of the new network interface and dynamic setup of new sub-queues to be assigned to the new network interface while the computer system remains in operation.

22. The computer system of claim 21, wherein the controller comprises a sub-queue setup engine configured to:

perform the dynamic setup of the new sub-queues when the new network interface is added to the computer system.

23. The computer system of claim 10, wherein the controller is configured to control the rate at which data is received by a sub-queue from the network interface associated with the sub-queue.

24. The computer system of claim 23, wherein the controller is configured to control the rate at which data is received by a sub-queue by using the occupancy rate of each sub-queue.

25. The computer system of claim 24, wherein the controller controls the rate at which message packets are received by each sub-queue by alternatively using polling and interrupt techniques to facilitate the transmission of message packets from the assigned network interface.